/[gzz]/manuscripts/Paper/paper.tex
ViewVC logotype

Diff of /manuscripts/Paper/paper.tex

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1.17 by jvk, Sat Mar 8 15:13:33 2003 UTC revision 1.18 by jvk, Tue Mar 11 13:14:52 2003 UTC
# Line 248  surface shape\cite{interrante97illustrat Line 248  surface shape\cite{interrante97illustrat
248  \subsection{Texture perception}  \subsection{Texture perception}
249    
250  Psychological studies on texture perception have mostly concentrated  Psychological studies on texture perception have mostly concentrated
251  on texture discrimination, the ability of human observers to discriminate  on \emph{texture discrimination}, the ability of human observers to
252  pairs of textures.  discriminate pairs of textures.  
253  % XXX: segregation vs. discrimination  The term is often used interchangably with \emph{texture segregation},
254    the more specific task of finding the border between areas of
255  First experiments on computer-generated, unnatural textures  different textures (different phases of local characteristics at the
256  in the 60s \cite{julesz62visualpattern} led to  border can segregate otherwise indiscriminable textures).
257  proposals of discrimination models based on  
258  $N$th order statistics of pixels and connectivity  First experiments on computer-generated, unnatural textures in the 60s
259  structures of certain micropatterns.  \cite{julesz62visualpattern} led to proposals of discrimination models
260    based on $N$th-order statistics (the joint distributions of all
261    $N$-tuples of pixels for given $N$) and connectivity structures of
262    certain micropatterns.
263    
264  Statistical modeling of textures as samples from a probability  Statistical modeling of textures as samples from a probability
265  distribution on a random field as already seen in \cite{julesz62visualpattern}  distribution on a random field as already seen in \cite{julesz62visualpattern}
266  in a simple form.  in a simple form.
267  The most popualar computational approach is Markov random fields  The most popualar computational approach is Markov random fields
268  \cite{cross83markov, geman84stochastic}, where a texture  \cite{cross83markov, geman84stochastic}, where the value of each pixel
269  is characterized by its local statistics.  depends only on the values of its neighborhood (local characteristics).
270  XXX: resolution-dependency?  XXX: resolution-dependency?
271    
272  Attempt to explain texture perception by the densities of textons  Attempt to explain texture perception by the densities of textons
# Line 280  Filtering based approach, e.g., \cite{he Line 283  Filtering based approach, e.g., \cite{he
283  Essentially a bank of linear filters is applied to the texture followed  Essentially a bank of linear filters is applied to the texture followed
284  by a nonlinearity and then another set of filters.  by a nonlinearity and then another set of filters.
285    
286    Mapping texture appearance to an Euclidian texture space
287    (see \cite{gurnsey01texturespace} and the references therein):
288    in the reported experiments, three dimensions have been sufficient
289    to explain most of the variation in the similarity judgements for
290    artificial textures.
291    However, the texture stimuli have been somewhat simple
292    (no color, lack of frequency-band interaction, etc.).
293    For some natural texture sets, three dimensions have also been
294    sufficient, but often the semantic connections cause the
295    similarity to be context-dependant, making it hard to assess the
296    dimensionality.
297    % XXX: this is something we should experiment with our textures
298    
299  XXX: reviews  XXX: reviews
300    
301  XXX: physiological knowledge of visual perception  XXX: physiological knowledge of visual perception

Legend:
Removed from v.1.17  
changed lines
  Added in v.1.18

savannah-hackers-public@gnu.org
ViewVC Help
Powered by ViewVC 1.1.26